Affect and Managerial Perfornnance: A Test of Hypotheses Barry...

Affect and ManagerialPerfornnance: A Test ofthe Sadder-but-Wiservs. Happier-and-SmarterHypotheses

Barry M. StawSigal G. BarsadeUniversity of California atBerkeley

This study provides a comparative test of twopsychological theories concerning the relationshipbetween affect and performance. Managerial simulationsare used to test whether people who are positive in ,disposition perform better or worse on both decisionaland interpersonal tasks. Results are consistent insupporting the happier-and-smarter as opposed to thesadder-but-wiser hypothesis, since they show positiverelationships between dispositional affect andperformance. The results are discussed in terms of theirrelevance to both the older literature on links betweensatisfaction and performance and the more recentcontroversy over the dispositional approach to jobattitudes.*

© 1993 by Cornell University.0001 -8392/93/3802-0304/$! .00.

We thank Jennifer Halpern for her role inthe original planning of this study and forher contributions to the development ofthe coding schenne used in this project.This study was nnade possible by afaculty research grant to the first authorfronn the Institute of Industrial Relationsat the University of California at Berkeley.

Managers have long believed that the happy worker is aproductive one. but decades of research on whether jobsatisfaction influences productivity have generally revealed aweak to nonsignificant relationship (Brayfield and Crockett.1955; Iffaldano and Muchinsky, 1985). Only turnover andabsenteeism have shown reliable linkages to satisfaction(Mobley. 1982). As a result of these findings, organizationalresearchers have had the unfortunate task of being thebearer of bad news, offering nnanagers consolations such as,"satisfaction may predict decisions to participate if notdecisions to perform" (March and Sinnon, 1958), or"satisfaction nnay follow from perfornnance if contingentreward systenns are in place" (Cherrington, Reitz, and Scott,1971). It is unknown whether these consolations have beenpersuasive or whether we have been successful inconvincing managers that job attitudes and perfornnance areloosely coupled. Nonetheless, most researchers have longrelegated the satisfaction-performance linkage to the folkloreof nnanagennent, as an unsubstantiated clainn of practitionersand the popular press (Staw, 1986).

Recently, sonnething interesting has happened that couldreopen the issue of whether people's attitudes andperfornnance are linked in organizations. While most of thefield has assumed that the attitude-perfornnance questionwas safely put to rest, the variables nnaking up thisrelationship have undergone nnetamorphosis (Staw, Sutton,and Pelled, 1993). No longer is job satisfaction the onlyoperationalization of attitudes at work. Instead, a number ofresearchers have been concerned with the expression ofemotion on the job (Rafaeli and Sutton, 1989), positive andnegative moods (Isen and Baron, 1991), and dispositionalaffect (Staw, Bell, and Clausen, 1986). Likewise, instead ofconsidering job perfornnance simply as a combination ofwork quantity and quality, other researchers have exploredextrarole behavior (O'Reilly and Chatnnan, 1986), citizenship(Organ, 1988), and task revision (Staw and Boettger, 1990).These expansions of the construct space for both attitudesand performance now make it possible to test new linkagesbetween these variables. In our view, one of the mostpronnising reformulations of the attitude-performancequestion involves the study of affect in organizations. Beforeposing theoretical arguments or hypotheses, however, it isnecessary to consider sonne definitional issues.

304/Administrative Science Quarterly, 38 (1993): 304-331

Affect and Performance

The Nature of Job Attitudes

Attitudes have historically been a broad construct used todenote cognitive, affective, and behavioral aspects of therelationship between a person and social, physical, orideological objects (e,g,, Katz and Stotland, 1959), Using thisdefinition, one's attitudes toward his or her job could includea set of beliefs about the work (e.g,. it is easy orchallenging), an affective reaction to the work (e,g,, one likesit or not), and behavioral intentions (e,g,, likelihood of leavingor recommending the job to a friend). An obvious problemwith this definition is that it is difficult to know where theattitudinal construct leaves off and behavior begins. As aresult, many researchers have followed Fishbein and Ajzen's(1975) recommendation that cognition, affect, and behaviorbe separated as much as possible and that attitudesprimarily reflect the affective component of the person-object relationship. Many organizational researchers havealso followed this convention. In an influential treatise on jobsatisfaction, Locke (1976: 1300), defined job satisfaction as"a pleasurable or positive emotional state resulting from theappraisal of one's job or job experiences," Smith, Kendall,and Hulin (1969: 6), in their development of the JDI indicatorof job satisfaction, similarly proposed that "job satisfactionsare feelings or affective responses to facets of thesituation,"

When job attitudes are explicitly equated with affectivestates, it is not much of a logical extension to argue that jobattitude research should be expanded by incorporating morevaried research on affect. This implies that job attitudesmean more than job satisfaction. Candidates for the study ofaffective reactions at work should therefore include therange of moods, emotions, and dispositions experienced bypeople in the organization. Whereas emotions, such as angeror fear, generally denote a strong reaction to a specificobject or cause, moods usually refer to a milder, morediffuse affective state that may not be directed toward anysingle attitudinal object (Lazarus, 1991), Likewise, anaffective disposition commonly refers to a general tendencyto experience a particular mood (e,g,, to be happy or sad) orto react to objects (e,g,, jobs, people) in a particular way(Lazarus, 1991), Thus, one might propose that job attituderesearch can be broadened by considering a wider range ofemotions directed toward the work situation, such as angerat supervisors, frustration with policies, or enthusiasm forthe product. Alternatively, the study of job attitudes could beenriched by delving more deeply into the affectivedispositions and moods that people bring to the worksituation, the route we follow in this paper,

A number of recent studies have tested the behavioralconsequences of affective states and dispositions (seeTaylor and Brown, 1988; Isen and Baron, 1991, for reviews).For example, affect has been related empirically to helpingbehavior (George and Brief, 1992), evaluation of others(Cardy and Dobbins, 1986), risk taking (Isen, Nygren, andAshby, 1988), negotiation skills (Carnevale and Isen, 1986),creativity (Isen, Daubman, and Nowicki, 1987). susceptibilityto influence (Schwarz. Bless, and Bohner, 1991). and biasesin information processing (Alloy and Abramson, 1988),

305/ASQ, June 1993

Certainly, as Isen and Baron (1991) noted, theseconsequences are relevant to our understanding of behaviorin organizational settings. But, because many of theseconsequences are conflicting or contradictory, it is not yetdear how affect influences the perfornnance of individualsworking on organizational tasks. In the study that follows,we therefore provide a comparative test of two fundamental,yet conflicting perspectives underlying the relationshipbetween affect and performance.

Affect and Managerial Performance

It can be argued that managerial performance is anespecially good arena for investigating the behavioralconsequences of affect. First, because the work ofmanagers is relatively unstructured, it may be highly subjectto influence by the person occupying the role (Miner, 1987;Bell and Staw, 1989), In contrast, because the tasks oflower-level employees are often machine-paced or highlyroutinized, it may be harder for them to control the way theyperform their jobs. Thus, we might expect that the affectivestates of managers would more likely be translated intobehavioral outcomes than would the affect of lower-levelworkers. Second, we believe that many of the behaviors onemight expect to be influenced by affective states, as derivedfrom previous psychological research, are central tomanagerial work (Mintzberg, 1973), Managers must makemany decisions, often with ambiguous and incomplete data,in order to cope with both daily demands and longer-termstrategic problems. Managers must also perform numerousinterpersonal tasks such as working with staff, supervisingothers, participating in meetings, and representing theorganization to outside constituencies. As we hypothesizebelow, individuals' affect may influence both thedecision-making and interpersonal aspects of managerialperformance.

Affect and Decision Making

A number of experimental studies have found links betweenpositive mood and decision-making performance. The bulkof this research, conducted by Alice Isen and her colleagues,has used mild positive mood inductions, such as beingoffered food or a small gift, to examine the cognitive andbehavioral consequences of affective states. Typical of thestudies in this literature is that by Isen and Means (1983)showing that people who were made to feel happyperformed a complex task (choosing a car for purchaseamong six alternatives, each differing along nine dimensions)more efficiently than control subjects. They were better ableto eliminate unimportant information and find usefulheuristics to solve the problem. Other research has indicatedthat positive affect can induce people to process informationor solve problems creatively, Isen and Daubman (1984)found that positive affect led subjects to use broadercategories for sorting information and to display moreflexibility in their categorization schemes. Likewise, Isen etal, (1985) showed that people who were happy, comparedwith controls, gave more unusual and more diverseassociations to neutral stimulus words. Finally, Isen,Daubman, and Nowicki (1987) demonstrated that increasing

306/ASQ, June 1993


In most of the research onsadder-but-wiser effects, students arecategorized as "depressed" on the basisof their questionnaire responses to aninventory such as the BDI (Beck et al.,1961). Most subjects characterized asdepressed in these studies are notseverely or clinically depressed but,instead, display a nnild or subclinical formof depression that could simply belabelled as unhappiness. Nearly one-thirdof college samples have typically beenplaced in this depressive group.

positive affect improved people's performance on Duncker's(1945) classic candle task and Mednick's (1962) RemoteAssociates Test, both commonly used measures ofcreativity. Thus, it appears that positive affect may serve asa stimulus for certain kinds of problem-solving—decisiontasks that are either so complicated that they requiresimplification and the use of heuristics, or activities that canbe aided by greater cognitive flexibility and originality inresponse.

Studies examining the consequences of longer-termaffective dispositions have also shown the beneficial resultsof positive disposition. Scheier, Weintraub, and Carver (1986)found that dispositional optimism (a generalized expectancyfor positive outcomes) was associated with better adaptationunder stress, problem-focused coping, seeking of socialsupport, and emphasis on the positive aspects of stressfulsituations. In contrast, dispositional pessimism wasassociated with denial, distancing, focus on stressfulfeelings, and disengagement from goals. Finally, in one ofthe most interesting tests of the behavioral consequences ofaffective disposition, Seligman and Schulman (1986) foundthat optimistic, as opposed to pessimistic, explanatory stylepredicted the success of life insurance agents. In a job inwhich persistence is essential for performance, those whowere dispositionally optimistic sold more insurance andstayed on the job longer than those who were dispositionallypessimistic.

From the psychological literature on positive mood andoptimism, one might hypothesize that there are distinctadvantages for managers who are positive in affect. Ingeneral, positive individuals appear to be more energized andalso able to direct their energies in more appropriatedirections than those who are negative in emotionality. Thus,when managers must make many decisions and cope withdecisions that are complex, involving competing andambiguous elements, positive affect shouid be a facilitatinginfluence. One could expect that the flexibility, creativity, andpersistence fostered by positive affective states will helpmanagers perform better on the decisional component oftheir work roles. Hence, we propose:

Hypothesis 1: Managerial decision nnaking will be facilitated orinnproved by positive affect.Sadder but wiser. The psychological literature on affectdoes not uniformly support the view that positive affect hasbeneficial consequences. Some research shows thatdepressed people may actually be more accurate in theirjudgments than their nondepressed counterparts, such thatthey are sadder but wiser.'' In a series of experimentalstudies. Alloy and Abramson (1979) found that nondepressedstudents significantly overestimated the degree ofcontingency between their responses and outcomes whenthe outcomes were desired and underestimated the degreeof contingency when the outcomes were undesired. Incontrast, depressed students were more accurate in theirjudgments regardless of whether the outcomes weredesired or not. A number of follow-up studies have generallyreplicated this depressive realism effect (e.g.. Alloy,Abramson, and Viscusi, 1981; Alloy and Abramson, 1982)

307/ASQ, June 1993

and have confirmed its causal direction (Martin, Abramson,and Alloy, 1984). Thus, although most people suffer from anoptimism bias (Matlin and Stang, 1978; Lichtenstein,Fischoff, and Phillips, 1982) and an illusion of control(Langer, 1975), those who are affectively negative seem tobe relatively immune to these shortcomings.

Studies supporting the sadder-but-wiser hypothesis indicatethat those with depressive tendencies may avoid a broadrange of self-protective biases (Taylor and Brown, 1988).They may be less biased in attributing the causes of positiveand negative outcomes (Sweeney, Anderson, and Bailey,1986), less likely to minimize the possibility of negativeevents, or to overestimate the chance of positive events(Alloy and Ahrens, 1987). Depressives may also be less likelyto perceive the world as under their control when it is,instead, subject to exogenous forces (Golin, Terrell, andJohnson, 1977). Finally, those who have depressivetendencies may be less likely to overestimate their ownabilities in ambiguous task situations (Tabachnik, Crocker,and Alloy, 1983). The implication of these data is that themost accurate information processing may be practiced bythose who are least positive in affect.

Additional support for the sadder-but-wiser hypothesis canbe deduced from the attitude-change literature. Recentresearch on attitude change has identified two differentinformation-processing strategies people may adopt whenthey are exposed to persuasive communications (Chaiken,1980; Petty and Cacioppo, 1986). It has been argued thatpeople use either systematic/central-route processing,involving detailed, analytic, and effortful evaluation ofmessages, or take a more heuristic/peripheral route,emphasizing surface features of the message or its source.Several empirical studies have shown that those who areinduced to be in positive moods are less likely to distinguishbetween high- and low-quality messages and more likely torely on source-credibility variables than those in neutral ornegative moods (see Mackie and Worth, 1991; Schwarz,Bless, and Bohner, 1991, for reviews). It can therefore beargued that positive mood may either reduce one's capacityfor information processing or inhibit one's motivation to gobeyond simple heuristic processing.

Because managerial roles often require effortful, systematic,and relatively unbiased processing of information, theperformance of individuals occupying these roles may beinfluenced by the cognitive illusions, biases, and simpleheuristic processing found in the depressive realism andpersuasion literatures. Thus, we can propose:

Hypothesis 2: Managerial decision making will be hindered bypositive affect.

Resolving the controversy. Before one can extend thepsychological research on affect to organizational settings, itis necessary to resolve the controversy between thesadder-but-wiser and what could be called the happier-and-smarter positions. To do this one cannot, however, simplysort managers according to their positive or negativeemotionality and correlate affective measures with ratings ofmanagerial performance. Organizational performance ratings

308/ASQ, June 1993


are too easily influenced by affective dinnensions, so thatone nnay be nneasuring positive halo rather thandecision-making skills (Cooper, 1981). A better solutionwould be for either current or future nnanagers to exhibittheir decision-making skills on an exercise that is designedto parallel realistic organizational problems. Such an exercisecould be objectively coded on specific dimensions that aredesigned to provide a fair test of the competing theoreticalpositions (Cooper and Richardson, 1986). For example, onemight argue that both the happier-and-smarter andsadder-but-wiser hypotheses would make diametricallyopposed predictions on the accuracy of decision making, aswell as information search, recognition of contingencies, andanalytical skills. The happier-and-smarter position would bethat positive affect is beneficial to each of these aspects ofdecision making, while the sadder-but-wiser position wouldpredict adverse effects on these same dimensions.

Affect and Interpersonal Behavior

Being positive in mood and disposition also appears toinfluence a broad range of interpersonal behavior, much of itself-reinforcing. As in a self-fulfilling cycle, positive peoplethink more positively of others (Gouaux, 1971; Baron, 1987),and others also find them to be more attractive (Coyne,1976; Cardy and Dobbins, 1986). The cyclicality of positiveaffect is probably due not only to possible halo effects(Cooper, 1981) but also to positive people actually beingmore interpersonally rewarding. A large literature links affectto helping behavior (e.g.. Aderman, 1972; Fried andBerkowitz, 1979; Cunningham et al., 1990). As Isen andBaron (1991) noted, positive affect leads to sociable andbenevolent acts, especially in situations in which such actshelp to maintain the individual's positive affective state (Isenand Simmonds, 1978). Positive affect also seems to beimportant in interpersonal conflict situations. Individuals withpositive affect have been found to be more cooperative andiikelier to find integrative solutions in negotiation contexts(Carnevale and Isen, 1986). Finally, positive people appear tohave more social influence. As Cialdini (1984) noted, beinglikable is useful for getting others to agree with one'sposition, so as to sell ideas as well as products. Perhaps, aspredicted by Petty and Cacioppo's (1986) attitude-changemodel, being positive puts others in a similar mood state,thus enabling the person to be persuasive via peripheral aswell as central-route processing of communication. Hence,both the weak and strong arguments of positive people maybe influential in their interpersonal dealings with others.

In terms of organizational behavior. Organ (1988) noted thatjob satisfaction is a predictor of citizenship behaviors, thosehelpful acts that go beyond the normal requirements of ajob. Similarly, George and Brief (1992) argued that positiveaffect is a broad determinant of spontaneous behavior inorganizations, encompassing acts such as helpingcoworkers, protecting the organization, making constructivesuggestions, and spreading goodwill. In many ways, bothOrgan (1988) and George and Brief (1992) have extendedthe consequences of affect beyond the person's interactionswith others to his or her relationship with the organization asa whole.

309/ASQ, June 1993

Our hypothesis will be limited to the interpersonal behaviorof managers with various levels of affect. We predict thatthose who are positive in affect will be more effective insituations in which interpersonal persuasion and negotiationare necessary—contexts in which managers must interactwith others to procure resources or get policies approvedand implemented. We predict that those who are mostpositive in affect will be rated most highly in terms of theirinteraction skills, participation in group decision making,leadership, and contribution to group solutions. Thus,

Hypothesis 3: Managers' interpersonal relations will be aided orimproved by positive affect.

It is difficult to derive a hypothesis that might compete withthe above position that happier individuals are morecooperative and interpersonally competent. One couldhypothesize that highly critical people are perceived to bemore intelligent than others because they have noticed theflaws in the arguments or preferences of others (Amabite,1983), Nonetheless, it is not clear how such perceptions ofintelligence would translate directly into improvements ininterpersonal relations, and as a result, any sadder-but-interpersonally-competent hypothesis must be regarded ashighly speculative.

Overview of the Study

The research reported here examined three hypothesesconcerning the relationship between affect and managerialperformance. Two of the hypotheses were contradictory,contrasting the sadder-but-wiser and happier-and-smartereffects on decision making, while the third hypothesis oninterpersonal behavior added a further test of thehappier-and-smarter position. As we describe below, adispositional measure of affect was related to performanceon both decisional and interpersonal exercises that weredesigned to simulate major aspects of managerialperformance. These data were supplemented by summaryratings of managerial potential for individuals of variousaffect levels. Together, these data help elucidate the role ofaffect in managerial performance,

METHOD

The data for this study come from the M,B,A, assessmentproject conducted by the Institute of Personality and SocialResearch (IPSR) at the University of California at Berkeley,This project is a joint effort of business and psychologyresearchers to understand the personality determinants oforganizational behavior.

Subjects. One hundred and eleven first-year M,B,A,students (64 men and 47 women) participated in anassessment center weekend. The mean age of the subjectswas 27,6 years, ranging from a low of 21 to a high of 40years of age. Subjects had an average of 3,4 years of workexperience since their undergraduate degrees.

Only 12 M,B,A,s participated in each weekend assessment.These students were assessed by a 12-member personalitystaff and a six-member managerial staff. The personalitystaff consisted of psychology faculty, researchers affiliated

310/ASQ. June 1993


with IPSR, and doctoral students enrolled in a practicumcourse on the assessment of personality. The managerialstaff consisted of faculty and doctoral students from theHaas School of Business and the organizational psychologyprogram at Berkeley, Although the psychology faculty andresearch affiliates of IPSR were already very experienced inassessment techniques, all members of the personality andmanagerial staff received extensive training for theassessment tasks. The weekend assessment consisted ofbehavioral exercises, a decision simulation, in-depthinterviews, and an extensive number of personalityinventories. The M,B,A, students received personalfeedback, as well as a small amount of course credit fortheir participation in the assessment.

Measurement of Affective Disposition

To measure affect we used a composite scale of positiveaffectivity. As defined by Watson (1988: 128), positiveaffectivity (PA) "reflects one's level of pleasurableengagement with the environment," High PA ischaracterized by high levels of enthusiasm, energy, mentalalertness, and determination, while low PA reflects lethargyand depression (Watson, 1988; Watson, Clark, and Carey,1988),

A scale of positive affectivity was constructed by averagingthe standardized z scores for the following three measures:(1) A single-item, 9-point self-report trait rating of positiveaffect. Positive affect was described as "cheerful,enthusiastic, optimistic, accentuates the positive, versuspessimistic, discouraged, and emphasizes the negative," (2)An average score of the personality assessors' rating of eachparticipant's positive affect on the same single-item 9-pointscale described above. As described, 12 personalityassessors rated each individual in terms of his or her affect.The mean interrater reliability was ,82, (3) The well-beingscale of the Multidimensional Personality Questionnaire(MPQ) (Tellegen, 1982), This scale consists of 24 true-falseitems assessing positive affect. Sample items are "I oftenfeel happy and satisfied for no particular reason," "It is easyfor me to become enthusiastic about things I am doing,"and "I live a very interesting life," Like other MPQmeasures, the well-being scale is factor analytically derivedand has been shown to have acceptable internal consistencyand reliability over time (Tellegen, 1982),

At least two of the three affect subscales had to be availablefor an M,B,A, participant to be included in the data set.Thus, the sample ranged from 94 to 111 subjects,depending on the dependent variable being analyzed. Theaverage interitem correlation of this three-part affect scalewas ,47, ranging from a low of a ,35 correlation betweenstaff ratings of affect and the well-being scale to a ,67correlation between the well-being scale and self-ratings ofaffect. The coefficient alpha for the three-part affect scalewas ,74,

Assessing Decision Making

Decision making was assessed through a managerialsimulation developed by Development Dimensions

3n/ASQ, June 1993

International (DDI) (1983). The simulation is a three-hourin-basket exercise that includes 21 different decision items.The format is one in which the subject must assume therole of plant manager, due to the sudden death of one'spredecessor. The subject must deal in written form withnumerous accumulated problems to ensure the plant runssmoothly. The exercise is designed to simulate both thecomplexity and ambiguity of actual managerial decisions.Some of the problems need quick action, while othersrequire the collection of additional data to make a sounddecision. Some of the problems can be addressed on theirown, while others require consideration of decisions onprevious or subsequent items. Several problems necessitateactions to be taken by the subject, while others require useof one's staff to gather data or carry out tasks. Among the21 decision items are problems involving manufacturing,personnel, research and development, and legal issues.

The original DDI coding scheme was revised for thisresearch. We developed specific and objective categories forbehavior on the in-basket, so as to reduce any problems ofcoder judgment and to increase replicability. The followingdecision dimensions were coded from the in-basketexercise: (1) accuracy: whether the decision made wasactually the correct one; (2) additional information: whetheradditional data was requested from subordinates or gatheredbefore making a decision; (3) situationai contingencies:whether subjects recognized the interrelationship ofproblems and decisions to each other; (4) use of data: useof quantitative reports such as productivity and absenteeismindexes in making decisions; (5) timeliness: whetherproblems that needed quick action were addressed andothers not undertaken prematurely; (6) delegation: whethertasks and decisions were delegated to the right person; and(7) follow-up: whether instructions were given to others forfuture reporting and meeting dates.

For each of the above coding categories, counts were madeon whether the participant displayed the particular orappropriate behavior. A one or zero was recorded for each ofthe decision items that provided an opportunity to displaythe coded behavior. When a decision item did not provide anopportunity to display the behavior, no coding was done. Atotal score was calculated for each participant for each ofthe decision dimensions.

There was no scoring for the time participants took tocomplete the DDI exercise, because every participantappeared to use all of the available three hours. Differencesin the quantity of items addressed and the thoroughness oftheir solution were assessed by the codings for decisionquality described above.

Of the dimensions described here, the first four constitutethe fairest comparative test of the sadder-but-wiser versusthe happier-and-smarter hypotheses, since for each of themone could theoretically predict opposite results. On the latterthree measures (timeliness, follow-up, and delegation), it isonly possible to draw one-way or unitary predictions. Forexample, for timeliness and follow-up it is possible to predictan adverse influence of positive affect, since the sadder-but-

312/ASa June 1993


wiser position would imply more careful timing and greaterfollow-up on the behavior of others. Likewise, one mightargue that positive people are better able to delegateproperly on a complex problem. These latter predictions aremore speculative inquiries, rather than comparative testsderived directly from the affect literature.

For each of the measured dimensions of decision making asimple count was used to note whether the behavior wasexhibited or not. We then calculated the number of items onwhich subjects exhibited a given behavior, out of the totalnumber of items on which there was an opportunity to showthe behavior (not all 21 items involved each of the aspectsof decision making). Finally, a percentage mean was notedfor each decision dimension.

In-basket exercises have been shown to be relatively validand reliable indicators of managerial decision making andadministrative skills (Frederiksen, 1966; Thornton andByham, 1982: 181-184). Such simulations may be especiallyuseful when one of the variables of a theoretical test is likelyto contaminate the measurement of performance. Severalexperimental studies have shown that rated positive affectand likability may lead to inflated performance ratings (Cardyand Dobbins, 1986; Krzystofiak, Cardy, and Newman, 1988;Smither, Collins, and Buda, 1989). Thus, to assessmanagerial decision making it may be advantageous to haveobjectively coded tasks such as in-basket exercises ratherthan in-situ performance ratings.

Assessing Interpersonal Performance

To assess interpersonal performance, we used a leaderlessgroup discussion (LGD) exercise. LGD exercises arefrequently used in managerial assessment centers and havebeen found to be both a valid and reliable measure ofinterpersonal skills and activity level (Bray and Grant, 1966;Thornton and Byham, 1982: 170-176). The LGD used in theM.B.A. assessment was a competitive exercise involving theallocation of limited resources. Each participant was asked torepresent a subordinate's interest in getting a pay raise, aswell as the best interests of the company. There wereneither sufficient resources to meet all requested raises norenough time to evaluate fully all the possibilities forallocating funds. Each participant not only had to persuadeothers of his or her claim on resources but also help movethe group toward an overall allocation scheme that wasacceptable to all group members. Thus, like manymanagerial tasks in which both distributive and integrativesolutions are possible, participants were asked to work fortheir own as well as joint interests.

M.B.A. students participated in the LGD in groups of six.Participants were given ten minutes to study the case andthen thirty minutes to reach a consensus decision on theallocation of pay raises for each of six candidates. Theparticipants were ranked by their peers (the other fivemembers of the group) on their overall performance andcontribution to group effectiveness. They were also rankedby 12 personality assessment staff members who observedthe group discussions. Staff ranked the followingdimensions: extent of participation, quality of participation,

313/ASQ, June 1993

politeness, leadership, criticality, social compliance, masteryof information, and task engagement. The mean staffreliability was .95; the mean peer reliability was .81 (see alsoJohn and Robins, 1991).

Other Measures

Managerial potential. M.B.A. students were also assessedon more global dimensions of management. As notedearlier, for each group of 12 participants there were sixmembers of a managerial assessment staff (separate fromthe 12-member personality staff used in the assessment ofpositive affect or the LGD exercises). The managerialassessors observed the LGD exercise (but did not fornnallyrate participants on it), read the participants' responses tothe in-basket exercise (without formally coding theresponses), and conducted in-depth interviews. At the endof all assessment activities, the managerial staff met inthree-person groups to discuss each participant'sperformance and, in turn, rated the participants on thefollowing 15 dimensions: initiative, stress tolerance, energy,leadership, sensitivity to others, analytic thinking, decisionmaking, creativity, factfinding, oral communication, writtencommunication, control, planning and organizing, anddelegation. The three-person assessment teams reachedconsensus ratings on each managerial dimension. Ifindividual staff ratings differed by no more than 1 point (on a5-point scale), the modal score was chosen as theconsensus rating. If any of the staff ratings differed by morethan 1 point, the assessors then discussed the participantand again rated that managerial dimension until consensuswas reached. These 14 consensus ratings were averaged tocreate a managerial dimensions scale (a = .90). In addition,a single-item rating scale of the overall managerial potentialof the participant was obtained from each assessment team.

Satisfaction. Though not part of the tests of thesadder-but-wiser or happier-and-smarter hypotheses, we alsomeasured the satisfaction of participants with the BerkeleyM.B.A. program. Because prior research on affectivedisposition has predicted job satisfaction (e.g., Staw, Bell,and Clausen, 1986), we believed it would be useful tocross-check these results with previous findings. Thus,satisfaction with the following 15 specific aspects of theM.B.A. program was assessed: course material, the qualityof teaching, fairness of grading, physical facilities of theschool, opportunities for interaction with faculty,opportunities for interaction with other students, size ofclasses, scheduling of classes, relevance of the program toone's career, opportunity to take desired courses, expertiseof the faculty, reputation of the school, amount of workrequired, opportunities to participate in school activities suchas clubs, and opportunities to participate in extracurricularactivities such as social events and parties. Seven-pointrating scales on each of these 15 items were summed toform an overall measure of satisfaction with the M.B.A.program (a = .75). An additional 5-item scale of general lifesatisfaction (including questions on the participant's sociallife, living situation, life in the Bay Area, progress in careerdevelopment, and overall quality of life) was alsoadministered to participants (a = .79).

314/ASQ, June 1993


Control variables. Because performance on the in-basket andLGD exercises could be affected by common demographicvariables, we controlled for sex. age. GMAT scores, andyears of post-undergraduate work experience. Ability andexperience might be expected to influence performance onthe in-basket, while age and sex could influence ability topersuade or lead others on the LGD task.

Analysis

Although the three hypotheses we tested were phrased interms of positive affect facilitating or hindering performance,our analyses are associative rather than causal. In addition, itis important to note that the theoretical literatures on whichthe hypotheses are based do not necessarily imply uniformrelationships across all levels of affect and performance. Theliterature supporting the happier-and-smarter effect, forexample, consists of studies comparing subjects withelevated levels of PA with control subjects who have notreceived any affect manipulation (e,g,, receiving a gift orfinding money in a telephone booth). Moreover, studiessupporting the sadder-but-wiser effect have generallycontrasted subjects who are at least mildly or subclinicallydepressed with those without depressive symptoms. Typicalof these studies is Alloy and Ahrens' (1987), in which 532college students were subdivided on the basis of self-reportmeasures of depression (Beck, 1967), Twenty-six percent ofthis sample was categorized as at least mildly depressed, 36percent as nondepressed, and the remaining students,between these extremes, were excluded from the study.Thus the theoretically appropriate test of the sadder-but-wiser position is an examination of the contrast betweenthose with high and low affect, rather than simply testing foran inverse relationship between affect and performance. Theconceptually parallel test for the happier-and-smarter positionis an examination of the contrasts between individuals withhigh positive affect and those with other affect levels.

In dividing a sample for theoretical contrasts it is ideal tohave measures scored exactly as they have been in theprevious literature (e,g,, using the same cut-off values on theBeck Depression Inventory), Because the IPSR data used acomposite scale of observational and self-report measures ofaffect, rather than the Beck inventory, we approximated thecut-ofif values of the depressive realism literature bytrichotomizing our sample on affect. We then conductedanalyses of covariance on the performance measures, usingsex, age, GMAT scores, and years of experience ascovariates. Those who were lowest in PA could thus becompared statistically with those who were highest in apriori tests of the sadder-but-wiser position, while those whowere highest in PA could be compared a priori with othersfor tests of the happier-and-smarter position. To examine thegenerality of the results beyond particular cut-off values foraffect, we also conducted hierarchical regression analyses inwhich performance was predicted by both the compositemeasure of affect and control variables,

RESULTSBecause of its conceptual and methodological fit with theprevious literature, analysis of covariance was the primary

315/ASQ. June 1993

procedure for comparing the sadder-but-wiser vs,happier-and-smarter hypotheses. All of the analysespresented here tested for the effects of high, medium, andiow positive affect on decisional and interpersonalperformance, using the control variables of sex, age, GMATscores, and years of experience as covariates. As across-check on the results we have included, in theAppendix, hierarchical regression analyses on bothdecision-making and interpersonal performance. For eachhierarchical regression analysis, control variables (sex, age,GMAT scores, and years of experience) were entered firstinto the equations predicting performance, followed by thecomposite measure of positive affect.

Table 1 reports the means and standard deviations of eachof the measures used in the study, as well as theirzero-order intercorrelations.

Decision Making

Table 2 shows the means for the high-, medium-, andlow-affect groups on each aspect of decision making, Amultivariate test examining whether there were differencesbetween affect groups, across all the decision-makingvariables, showed a highly significant effect, [F(14,194) =2,39, p < ,005], A univariate analysis of covariance showedsignificant differences between affect groups in overallaccuracy or correctness of decisions [F(2,104) = 3.12, p <,05], There also were significant univariate effects on thethree decision-process measures of information requests

Table 1

Means, Standard Deviations, and Intercorreiations of iVIeasures

Variables

1, GMAT2, Age3, Gender (0 = male, 1 = female)4, Years of work experience5, Affect6, Decision-making accuracy7, Amount of additional information requested8, Recognition of situational contingencies9, Use of quantitative indices

10, Analytic index11, Timeliness in decision making12, Correct decision delegation13, Amount of follow-up14, Peer rankings of performance and

contribution to group effectiveness in LGD15, Ouality of participation ranking in LGD16, Mastery of information ranking In LGD17, Leadership ranking in LGD18, Extent of participation ranking in LGD19, Task engagement ranking in LGD20, Being critical ranking in LGD21, Social compliance ranking in LGD22, Politeness ranking in LGD23, Managerial dimension scale ratings24, Overall rating of managerial potential25, Satisfaction with the MBA program26, General life satisfaction

/V»

111111111111111111111111111111111111111

839494949494949494

111111111111

Mean

625,0527,55

,423,38

,00,56,31,48

- ,01,01,86,62,38

3,436,636,536,546,536,526,546,536,543,172,544,605,30

S,D.

68,503,87

,502,94

,81,18,14,30,77,64,11,13,15

1,332,432,332,742,882,272,542,662,35

,59,87,63

1,06

* The sample size for an intercorrelation will be that of the lower number of

1

,20,02,06

- ,05,25

- ,08,25,33,24,15,35,08

,19,29,31,24,20,27,14

- ,06,04,26,35,05

- ,12

Zero-order correlationst

2

,10,59

- ,30,03,19,14,32,31,12,17,12

,11,20,24,14,13,18,09

- ,08,03,17,16

- ,19- ,17

the pair.

3

- ,03,02,14,13,22,25,28,02,20,19

,01,11,12,06,03,15

- ,09,06,23,25,08

- ,23,03

4

- ,04,02,26,03,14,20,05,01,14

,15,13,17,12,17,14,14

- ,14- ,09

,13,18

- ,11- ,11

5

,20,16,06,03,12

- ,05,06

- ,12

,20,24,17,24,24,18,00

- ,02,10,15,20,30,52

6

,07,25,35,31,39,45,11

,14,21,20,18,13,17,02,00,12,37,23,17,12

7

,19,17,63,15

- ,05,31

,16,18,18,17,13,16,04

- ,07,06,18,08

- ,12,12

t Correlations above ,19, p < ,05, with a sample size of 111; correlations above ,20, p < ,05, with a sample size of 94; correlations above,22, p < ,05, with a sampie size of 83; all two-tailed tests.

316/ASQ, June 1993


[F(2,104) = 3.17, p < .05], recognition of situationaicontingencies [F(2,104) = 3.06, p < .05], and use of data[F(2,104) = 3.35, p < .05]. Combining these three processmeasures into an overall analytic skills index yielded a highlysignificant effect [F(2,104) = 4.84, p < .01]. Each of thesestatistical effects was in the direction of facilitation,supporting the happier-and-smarter hypothesis, usingtwo-tailed tests of significance. There were no significanteffects for the timeliness of decisions, decision delegation,or the amount of follow-up requests.

Table 2 shows the a priori contrasts used for making specifictests of the sadder-but-wiser and happier-and-smarterpositions. In no case did an a priori contrast support thedepressive realism effect; the low-affect group wasinvariably below the high-affect group in mean performanceon the decision task. The predominant effect was in thedirection of a facilitating influence of affect, though thedifferences between the various means did not appear as aperfectly linear trend. Regression analyses using polynomialmodels did not, however, reveal significant curvilinearity inthe data.

Regression analyses on decision making, shown in Table A.Iin the Appendix, yielded similar, though weaker results thanthose of the analyses of covariance. After controlling for theeffects of GMAT, age, gender, and years of experience, theregression equations showed that affect significantlypredicted decision-making accuracy, the amount ofinformation requested, and the analytical index. Unlike the

Table 1 (continued)

10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

3875033012

12151516111001020318202209

.68

.19

.35

.13

.24

.18

.25

.14

.12

.16

.09-.01.06.31.25

-.07-.03

.17

.29

.27

.24

.24

.27

.23

.17

.20

.06-.06.07.30.24

-.20.09

.26

.12

.12

.22

.19

.21

.20

.24

.11-.09.01.29.21.12.06

.18

.09

.19

.25

.19

.12

.16-.03.08.12.33.28

-.03.01

-.05-.03.06

-.02-.08.05

-.13.09.17.08.02

-.06-.11

.8011.86.81.75.52

-.48-.11.51.49.03.06

.90

.91

.76

.87

.46-.42.06.55.49.02.10

.89

.84

.90

.63-.61-.17.54.48.01.09

.91

.87 .85

.62 .79 .62- . 6 0 - .77 - .61 - .93- . 18 - . 44 - . 13 - . 78 .79

.57 .52 .52 .36 - . 2 8 - .01

.54 .51 .49 .37 - . 31 - .12 .75

.01 - . 04 - .01 - . 06 .05 .10 .04 .20

.09 .08 .12 .02 - . 07 .00 .02 .07 .37

317/ASQ, June 1993

analysis of covariance results, however, the regressioncoefficients did not indicate a significant influence of affecton recognition of situational contingencies or the use ofquantitative indices.

Interpersonal Performance

Table 3 shows the interpersonal perfornnance of individualsdivided according to affect group. A nnultivariate testexannining differences across all the interpersonal variablesshowed a highly significant effect [F(18,124) = 2.16, p <.01]. A univariate analysis of covariance showed significantdifferences between the affect groups on peer ratings ofperfornnance [F(2,76) = 3.61, p < .05]. There were alsosignificant univariate effects on observers' ratings of qualityof participation [F(2,87) = 4.44, p < .05], nnastery ofinfornnation [F(2,87) = 3.65, p < .05] and leadership [F(2,87)= 3.52, p < .05], as well as a marginally significant effecton the extent of participation [F(2,87) = 2.47, p < .10]. Eachof these effects, as well as the a priori contrasts shown inTable 3, indicated that more positive individuals performedbetter on the interpersonal task. There were no significanteffects on the interpersonal dimensions that were lessdirectly related to performance, such as task engagement,being critical, social compliance, and politeness.

The regression results on interpersonal performance, shownin Table A.2 of the Appendix, were similar to those of theanalyses of covariance, although somewhat stronger. Aftercontrolling for effects of GMAT, age, gender, andexperience, there was a significant influence of affect on

Table 2

Analysis of Variance of Affect

Decision making

Decision-making accuracy

Amount ot additional informationrequested before makingdecisions

Recognition of situationalcontingencies

Use of quantitative indices indecision making (e.g..productivity, absenteeism, andattitude indices—mean ofnormalized measures)

Analytic index (index ofadditional informationrequested prior to makingdecisions, situationalcontingencies recognized, anduse of quantitativeindices—mean of normalizedmeasures)

Timeliness in decision making

Correct decision delegation

Amount of follow-up requested

• p < .05; - p < .01.

: and Decision iVflaking*

Overallmean

.557(.183).313

(.141)

.478(.297)

-.013(.772)

.008(.644)

.855(.105).620

(.127).379

(.147)

Low-affectgroupmean

.496(.176).266

(.146)

.431(.305)

- .04(.871)

-.162(.683)

.864(.09).589

(.133).374

(.149)

Med.-affectgroupmean

.581(.176).334

(.122)

.428(.276)

-.216(.648)

-.069(.556)

.854(.114).637

(.122).413

(.163)

• Controlling for GMAT, age, gender, and years of experience. Adjustedmedium-, and high-affect groups (111 total)

High-affectgroupmean

.594(.185).340(.149)

.575(.29).22

(.772)

.256(.64)

.847(.110).635

(.126).350

(.124)

OverallF-test

3.12*

3.17*

3.06*

3.70*

4 .94-

.21

1.85

1.68

means are reported. N =

Groupcomparison

(p < .05)

Lov\/ vs. highLow vs. mediumLow vs. highLow vs. medium

Low vs. highMedium vs. highMedium vs. high

Low vs. highMedium vs. high

37 for the low-.

318/ASO, June 1993


Table 3

Analysis of Variance of Affect and Leaderless Group Discussion Performance Rankings*

Leaderless groupdiscussion performancerankingst

Peer ranking of overallperformance & contribution togroup effectiveness (1-6)*

Quality of participation(1-12)Mastery of information(1-12)Leadership(1-12)Extent of participation(1-12)Task engagement(1-12)Being critical(1-12)Social compliance(1-12)Politeness(1-12)

Overallmean

3.43(1.33)

6.53(2.43)6.53(2.33)6.54(2.74)6.53(2.88)6.52(2.27)6.54(2.54)6.53(2.56)6.54(2.35)

Low-affectgroup mean

3.31(1.29)

5.89(2.26)6.21(2.19)5.90(2.8)5.79(2.88)6.33(2.32)6.69

• (2.6)6.34(2.69)6.13(2.29)

Med.-affectgroup mean

3.04(1.29)

6.13(2.47)5.96(2.61)6.09(2.85)6.32(3.2)6.11(2.47)6.38(2.89)6.84(2.69)6.43(2.08)

High-affectgroup mean

3.98(1.23)

7.50(2.31)7.35(1.93)7.55(2.31)7.39(2.37)7.07(1.95)6.54(2.14)6.41(2.33)7.02(2.62)

OverallF-tesl

3 . 6 1 -

4.44-

3.65-

3.52-

2.47*

1.62

.11

.31

1.12

Groupcomparison

(p < .05)

Low vs. high!Medium vs. high

Low vs. highMedium vs. highLow vs. highMedium vs. highLow vs. highMedium vs. highLow vs. high

•p< .10; -p< .05.• Controlling for GMAT. age. gender, and years of experience. Adjusted means are reported; N = 30 for the low-affect

group. N = 31 for the medium-affect group, and N = 33 for the high-affect group (94 total).t Variables were recoded so that the higher the number, the higher the ranking.* N = 30 for the low-affect group. N = 27 for the medium-affect group, and W = 26 for the high-affect group (83 total).§ p < .10.

ratings of performance, participation, nnastery of infornnation,leadership, extent of participation, and task engagennent.

Ratings of Managerial Performance

Table 4 shows the cell means for the three affect groups onstaff ratings of managerial performance. There weresignificant effects on both the managerial dimensions scale[F(2,104) = 3.68, p < .05] and the overall rating ofmanagerial performance [F(2,104) = 4.25, p < .05].Consistent with the data on decision making andinterpersonal performance, those who were positive in affectwere rated higher in managerial potential than those lower inaffect. The regression results, shown in Table A.3 in theAppendix, also showed significant effects of affect on

Table 4

Analysis of Variance of Affect and Ratings on Managerial Dimensions*

Managerial dimensions

Managerial dimension scale

Overall managerial potential

Overallmean

3.168(.594)2.541(.872)

Low-affectgroup mean

2.960(.594)2.216(.863)

Med.-affectgroup mean

3.269(.564)2.702(.861)

• p < .05.* Controlling for GMAT. age. gender, and years of experience. Adjusted

medium-, and high-affect groups (111 total).

High-affectgroup mean

3.274(.591)2.703(.830)

OverallF-iesl

3.68*

4.25*

means are reported. N

Groupcomparison

(p < .05)

Low vs. highLow vs. mediumLow vs. highLow vs. medium

= 37 for the low-.

319/ASa,June 1993

ratings of managerial performance, after controlling forGMAT. age. gender, and years of experience,

DISCUSSION

It is often difficult, if not impossible, to build a perfectcomparative test of theories. Advocates of each theoreticalposition may argue that the operationalization and testingfavors one particular approach or another. Though this studywas not immune to these potential difficulties, we wouldargue that both the sadder-but-wiser and happier-and-smarter positions were given a fair chance for confirmationby this research. The results of the comparative tests werenot of uniform strength, but they were consistent indirection. On each set of dependent variables—those ofdecision making, interpersonal performance, and ratings ofmanagerial potential—there was at least some evidence of afacilitating rather than an inhibiting role for positive affect.

Decision Making

The high-PA group was more accurate in its decisions,getting more of the in-basket items correct than those ineither the low or mid-level affect groups, High-PA peoplealso seemed to perform better on the processes underlyinggood decision making. They requested more informationwhen there were insufficient data to make reasonabledecisions. And. at least in the covariance analyses. high-PApeople had a greater tendency to recognize situationalcontingencies, such as seeing when one decision wouldadversely influence another, and to make greater use of thedata provided in the exercise. These significant relationshipsconsistently supported the happier-and-smarter rather thanthe sadder-but-wiser position.

There were no significant effects on timeliness, delegation,and follow-ups requested. As noted earlier, these measureswere not designed as comparative tests of the two theories.Although we predicted that low-PA people would be morecareful in their timing of decisions and tend to retaindecisions for themselves, because such predictions werenot directly implied by the depressive realism literature, it isperhaps understandable why significant results were notobtained. In contrast, it could be argued that the follow-upmeasure was a stronger operationalization of careful andsystematic decision making. This was the only direct test ofthe sadder-but-wiser position that did not show significantresults in the direction opposite to prediction.

The accuracy and care that high-PA people displayed in theirdecision making was precisely counter to what would havebeen expected had individuals used more heuristic ratherthan systematic processing of information (Chaiken. 1980),Perhaps managerial decisions cue M,B,A, students into aheuristic of careful decision making. Such an argumentwould, of course, be stretching the intention of thesadder-but-wiser literature and make its hypothesis virtuallyuntestable. because nearly any result could confirm thetheory. More likely is the explanation that positive affectprovides an energizing function that enables people to delvemore deeply into decision tasks. Positive individuals tend, asposited by Isen and Baron (1991). to search more broadly

320/ASQ. June 1993


and to consider information more thoroughly than those whomust labor with unhappy thoughts or depressed affect.

Interpersonal Behavior

The leaderless group data also showed a positiverelationship between affect and interpersonal performance.On both peer and staff ratings, people who were high inpositive affect scored better on several measures ofinterpersonal behavior. Positive individuals were judged bytheir peers to have made greater contributions to groupeffectiveness. Their participation and leadership were alsorated more highly by staff observers, and so was theirmastery of information needed to present their casespersuasively. Interestingly, there were no significant effectson the measures of being critical, social compliance, andpoliteness. What this pattern of data seems to show is thatpositive affect is not necessarily responsible for people beingjust nice or accommodating but, rather, more effectiveinterpersonally. This pattern of results also runs counter tothe possibility that the data are simply a function of haloeffects. If halo or liking accounted for all the interpersonaleffects, one would have expected the strongest results tohave been manifested on affectively toned items. Yet, therewere no significant effects on the items such as beingcritical, social compliance, and politeness. Instead, thestrongest results were evidenced on the performance-related measures, such as extent and quality of participation,mastery of information, and contribution to groupeffectiveness. Moreover, if the results were simply aproduct of halo, one might have expected that the effects onpeer ratings would have been stronger than those on ratingsby trained staff. Yet, as shown in Table 2, there were nosignificant differences between staff and peer ratings in therelationship between affect and interpersonal performance.

What the interpersonal data show is that people high inpositive affect are more competent interpersonally, able tocontribute more to group solutions and exert a leadershiprole among others. In many ways, these results should notbe surprising, for positive affect has been found to beassociated with extraversion and sociability (Tellegen, 1982),Nonetheless, these results may also indicate that high-PAindividuals are simply competent people who not only canmake good decisions on solitary tasks such as the DDI butcan also influence a group in productive ways.

Managerial Potential

The most global measures used in this study were ratings ofthe managerial potential of participants. Much as in anassessment center (Thornton and Byham, 1982), managerialstaff used observations and interviews of the individuals, aswell as the DDI and LGD exercises to judge the careerpotential of each participant. Though not independent of thesimulation data, the assessments were derived from adifferent set of judges and based on a broader coding ofperformance than the measures of decision making andinterpersonal behavior. If one assumes that the managerialstaff (business school faculty and doctoral students) aregood judges of future achievement, then these ratings meanthat those with positive affect are more likely to be

321/ASQ, June 1993

successful in their careers. If one is skeptical about theability of assessment centers to predict performance, andperhaps considers these ratings to be biased by halo effects,then one must still confront the fact that high-PA individualsare more likely to create positive impressions underlyingjudgments of managerial potential.

In organizations, managers' salary and promotion decisionsare seldom a simple averaging of objective achievements.Except in sales positions, quantitative data are rarelyavailable for judging the performance of managers, and evenwhen clearcut data are present, they must often beconditioned by qualitative factors, such as a major plantclosing in the person's sales district. Thus mostorganizational evaluations are based on an amalgam ofquantitative and qualitative information, not unlike thatavailable to our managerial staff, whose task it was to rateeach participant's potential. As a result, we would expectthe managerial staff ratings to be a reasonable prognosis ofhow this sample will fare in organizations. From theseresults we can conclude that high-PA people are more likelyto be rated positively by their organizations and be moresuccessful in their work careers than low-PA people.

Fit with Concurrent Research

Two recent reanalyses of field data completed after the startof this research showed conceptually parallel results. Usingavailable longitudinal data, Staw, Sutton, and Pelled (1993)found that positive affect was a significant predictor ofincreases in supervisory evaluations, pay, and social supportat work. House, Howard, and Walker (1991) likewise foundthat a composite measure of optimism predicted promotionsover time among a sample of AT&T managers. These fieldstudies add external validity not available from our simulationdata using the M.B.A. sample. At the same time, thesystematic observations provided by our decision-makingand interpersonal tasks add a degree of precision andinternal validity not present in these archival analyses of fielddata. Because the evaluation of performance in organizationscan be subject to so many influences, such as sociallabeling, liking, power, and social similarity, it is important toknow if positive affect really is related to effectiveness onmanagerial tasks. The present data on decisional andinterpersonal performance strongly suggest this to be thecase.

Attitudes and Performance

We noted at the outset of this paper that attitude-performance research could be improved by a broaderoperationalization of the constructs and that one promisingavenue was to expand attitudinal research beyond thetraditional measurement of job satisfaction to theinvestigation of affect in organizations. Measures of affectand satisfaction are no doubt related (e.g., in this study,affect was correlated .45 with the 5-item scale of lifesatisfaction and .30 with the 15-item scale of M.B.A.program satisfaction). In addition, dispositional affect hasbeen shown to significantly predict job satisfaction overpeople's entire working careers (Staw, Bell, and Clausen,1986). Nonetheless, by focusing on affect rather than

322/ASQ, June 1993


satisfaction it may be possible to be nnore successful inpredicting organizational performance.

Dispositional affect may be a more useful predictor oforganizational performance than satisfaction because it ismore stable and enduring over time, allowing continual asopposed to fleeting attitudinal influences on behavior. Forexample, the person who tends to be optimistic and viewsthe world positively may generally be more persistent andenergetic in work contexts (Seligman and Schulman, 1986).The positive individual may also approach customers in amore friendly and engaging manner (Rafaeli and Sutton,1989). These tendencies may not be as dependent on thetransient mood of the individual, the particular conditions ofemployment, or social comparisons often underlying jobsatisfaction and dissatisfaction. Measures of dispositionalaffect might especially have advantages in predicting theperformance of employees over long stretches of time.Performance ratings, as well as pay and promotiondecisions, are infrequent events designed to reflect months(or years) of previous employment. Therefore, if the effectsof job satisfaction are ephemeral, we would not expect thereto be as strong a relationship between satisfaction andperformance as the linkage between dispositional affect andperformance.

Positive affect may also be a predictor of performancebecause it is a repository of past experiences with theworld. As Campbell (1963) noted 30 years ago, attitudesshould be considered as acquired behavioral dispositions,where one's orientation toward an attitudinal object isconsidered to be a product of past relationships as well as aguide to future interactions. Assuming that prior success isreinforcing, one might therefore expect that successfulpeople would be high in dispositional affect. A positive andoptimistic view of events and people implies that the worldhas been and is likely to remain benevolent. For example.Dunning and Story (1991) recently found that students whowere categorized as depressed predicted and experiencedmore negative events than those who were not depressed.It is, of course, arguable whether positive affect is a causeof success, such that positive people do things to improvetheir chances in life, or, alternatively, whether affect issimply a consequence of one's performance. Our data, alongwith those of Seligman and Schulman (1986), imply thatpositive people do perform tasks differently than those whoare negative, although an answer to the exact causalrelationships between affect and performance must awaitfurther research.

Logically, one would expect the relationship between affectand performance to be dependent on the type of taskinvolved. Perhaps both the decision simulation and LGDexercise used here were more receptive to affectiveinfluences than tasks used in typical performance studies.For example, having to work on a repetitive assembly line,without the need for complex cognitive processing, mightdiminish or reverse the advantages of a positive disposition.It is also possible that other tasks may be more receptive tosadder-but-wiser effects. Low-PA people might do better ona forecasting or risk-taking task than those with high PA. The

323/ASQ, June 1993

low-PA person may also have special advantages inperforming a job in which negative affective displays arenecessary, such as that of a bill collector (Sutton, 1991).Sorting out these possible interactions between affect andtask type should be part of the agenda for future researchon job attitudes.

Some Measurement Considerations

It is conceptually possible, as we noted earlier, to breakattitudes into cognitive and affective components. Inpractice, however, it is difficult to draw firm boundariesbetween the affective and cognitive realms. Most attitudescales contain both affective and cognitive items (Brief andRoberson, 1989), and even measures of positive affectusually contain some cognitive items. Though it may becleaner theoretically to use purely evaluative measures totap affective disposition, the incorporation of cognitive items(usually those on optimism/pessimism) may increase thepower of prediction. Thus, comparative studies using bothaffective (e.g., Tellegen's well-being scale or Watson'sPANAS measure) and cognitively based measures (e.g.,Seligman's measure of explanatory style) may be a usefulavenue for future research.

There is still some controversy over whether affectrepresents a single dimension, anchored by positive-negativepoles, or two separate dimensions in which positive affect isnot the polar opposite of negative affect. Watson and hisassociates have strongly argued the case for treatingpositive and negative affect as separate conceptualdimensions (Watson and Tellegen, 1985; Watson, Clark, andCarey, 1988). As both Watson (1988) and Tellegen (1985)pointed out, however, the dimensionality of affect seems tobe highly dependent on the adjective descriptors used in thescales. They noted that many mood terms reflectinghappiness ("happy, cheerful") and contentment ("satisfied,pleased, contented") reflect tDoth high positive affect and lownegative affect. Conversely, descriptors denoting depression("sad, blue, gloomy") and loneliness ("alone, lonely") serveas significant markers for both low positive affect and highnegative affect. Watson and Tellegen (1985) labeled thesedescriptors "pleasantness terms" and cautioned againstusing these mixed markers in affect scales. To us, thesepleasantness items, avoided by some of the popular affectscales such as the PANAS, may be exactly the descriptorsneeded to best capture the construct of affective disposition.In conducting future research, we therefore believe it isappropriate to measure happy and sad states, regardless ofwhether they are incorporated into a single bipolar scale(e.g., Diener and Emmons, 1984) or two scales measuringpositive and negative affect.

Further Implications of Dispositional Affect

Past research has shown that happiness should not beconsidered simply as a transient state. Though situationalconditions can certainly influence one's mood or reactions toa particular stimulus, it has been shown that affectivedisposition has continuity over time (Costa, McCrae, andZonderman, 1987) and perhaps even a genetic component(Arvey et al., 1989; Bouchard and McGue, 1990). The

324/ASO, June 1993


implications of thinking of happiness as a dispositionalvariable are substantial. One consequence is that jobattitudes may be as dependent on the disposition of theindividual as on the characteristics of the job or workingconditions (Staw and Ross. 1985), Though controversial (seeDavis-Blake and Pfeffer. 1989). some evidence already existsto support the influence of dispositional affect on jobsatisfaction (e,g,. Staw. Bell, and Clausen. 1986),

A second implication of the dispositional nature of affectconcerns the selection of organizational members. Staw andRoss (1985) had previously cautioned against the selectionof employees on the basis of affective disposition. Eventhough dispositionally positive people are likely to havegreater job satisfaction, they noted, based on Alloy andAbrahamson's (1979) findings, that the most positiveindividuals may not be the best performers. They arguedthat roles involving critical and decision-making skills wouldbe most subject to a depressive realism effect, thuscanceling any benefits that dispositionally positiveemployees might bring in terms of job satisfaction andrelated participation behaviors. After completing this study,the costs of positive affect are yet to be discovered. Positivepeople may instead be more productive, at least on thedecision-making and interpersonal tasks assessed in thisstudy. Additional research is of course necessary to uncoverany overlooked costs as well as additional benefits ofpositive affect in organizations.

Conclusion

We began this article by noting that managers have longbelieved that the happy worker is a productive one. eventhough decades of data collection have failed to substantiatesuch a claim. What this study has shown is that it may betime to reopen the happiness-productivity question. Our datashow that managers' deep-seated beliefs may not befundamentally wrong, only off-base in terms of the specificmechanisms involved, Happy people may indeed be thosewho are most productive; however, rather than happinessand productivity being related as two separate, transient

, states, where increased satisfaction presumably results inincreased performance, it is possible that such a relationshipis more dispositionally based. We have shown that thosewho are high in dispositional affect may also perform betteron tasks that involve decision making and interpersonalrelations. It remains for future research to sort out the causaldynamics involved. We do not yet know whetherdispositional affect triggers more transient affective states,which in turn influence performance; whether affect andperformance are both part of a higher-order syndrome ofpersonality traits; or whether both affect and performancecovary over time with enduring features of people's workand life situation. Each of these inquiries has merit as apossible route for future research. Thus, we would arguethat not only has belief in the happy-productive workerprobably been dismissed prematurely, but that thisseemingly simple hypothesis can and should be transformedinto a series of more interesting and complex researchquestions,

325/ASQ. June 1993

REFERENCES

Aderman, David1972 "Elation, depression, and

helping behavior," Journal ofPersonality and SocialPsychology, 24: 91-101,

Alloy, Lauren B., and Lyn Y.Abramson1979 "Judgment of contingency in

depressed and nondepressedstudents: Sadder but wiser?"Journal of ExperimentalPsychology: General, 108:441-485,

1982 "Learned helplessness,depression and the illusion ofcontrol," Journal ofPersonality and SocialPsychology, 42: 1114-1126,

1988 "Depressive realism: Fourtheoretical perspectives," InLauren~ B, Alloy (ed,).Cognitive Process inDepression: 223-265, NewYork: Guilford,

Alloy, Lauren B., Lyn Y.Abramson, and Donald Viscusi1981 "Induced mood and the

illusion of control," Journal ofPersonality and SocialPsychology, 41: 1129-1140,

Alloy, Lauren B., and Anthony H.Ahrens1987 "Depression and pessimism

for the future: Biased use ofstatistically relevantinformation in predictions forself versus others," Journal ofPersonality and SocialPsychology, 52: 366-378,

Amabile, Teresa M.1983 "Brilliant but cruel:

Perceptions of negativeevaluators," Journal ofExperimental SocialPsychology, 19: 146-156,

Arvey, Richard D., Thomas J.Bouchard, Nancy L. Segal, andLauren M, Abraham1989 "Job satisfaction:

Environmental and geneticcomponents," Journal ofApplied Psychology, 74:187-192,

Baron, Robert A.1987 "Mood interviewer and the

evaluation of job candidates,"Journal of Applied SocialPsychology, 17: 911-926,

Beck, Aaron T.1967 Depression: Clinical,

Experimental and TheoreticalAspects, New York: Harper &Row,

Beck, Aaron T., C. H. Ward, M.Mendelson, J, E. Mock, and J.Erbaugh1961 "An inventory for measuring

depression," Archives ofGeneral Psychiatry, 4:561-571,

Bell, Nancy E., and Barry M. Staw1989 "People as sculptors versus

sculptor: The roles ofpersonality and personalcontrol in organizations," InMichael B, Arthur, Douglas T,Hall, and Barbara S, Lawrence(eds,). Handbook of CareerTheory: 232-251, New York:Cambridge University Press,

Bouchard, Thomas I., andMatthew McGue1990 "Genetic and rearing

environmental influences onadult personality: An analysisof adopted twins rearedapart," Journal of Personality,58: 263-292,

Bray, Douglas W., and Donald L.Grant1966 "The assessment center in the

measurement of potential forbusiness management,"Psychological Monographs:General and Applied, 80: 1-27,

Brayfield, Arthur H., and Walter H.Crockett1955 "Employee attitudes and

employee performance,"Psychological Bulletin, 62:396-424,

Brief, Arthur P., and LoriannRoberson1989 "Job attitude organization: An

exploratory study," Journal ofApplied Social Psychology, 19:717-727,

Campbell, Donald T,1963 "Social attitudes and other

acquired behavioraldispositions," In Sigmund Koch(ed,). Psychology: A Study of aScience, 6: 94-172, New York:McGraw-Hill,

Cardy, Robert L., and Gregory H.Dobbins1986 "Affect and appraisal accuracy:

Liking as an integral dimensionin evaluating performance,"Journal of Applied Psychology,71: 672-678,

Carnevale, Peter J, D., and AliceM. Isen1986 "The influence of positive

affect and visual access onthe discovery of integrativesolutions in bilateralnegotiation," OrganizationalBehavior and Human DecisionProcesses, 37: 1-13,

Chaiken, Shelly1980 "Heuristic versus systematic

information processing andthe use of source versusmessage cues in persuasion,"Journal of Personality andSocial Psychology, 39:752-766,

Cherrington, David J., H. JosephReitz, and William E. Scott1971 "Effects of contingent and

non-contingent rewards onthe relationship betweensatisfaction andperformance," Journal ofApplied Psychology, 56:531-536,

Cialdini, Robert B.1984 Influence: How and Why

People Agree to Things, NewYork: Morrow,

Cooper, William H.1981 "Ubiquitous halo,"

Psychological Bulletin, 90:218-244,

Cooper, William H., and Alan J.Richardson1986 "Unfair comparisons," Journal

of Applied Psychology, 71:179-184,

Costa, Paul T,, Robert R. McCrae,and Alan B. Zonderman1987 "Environmental and

dispositional influences onwell-being: Longitudinalfollow-up of an Americannational sample," BritishJournal of Psychology, 78:299-306,

Coyne, James C,1976 "Depression and the response

of others," Journal ofAbnormal Psychology, 85:186-193,

Cunningham, Michael R,, David R.Shaffer, Anita P. Barbee, PatriciaL, Wolff, and David J. Kelley1990 "Separate processes in the

relation of elation anddepression to helping: Socialversus personal concerns,"Journal of Experimental SocialPsychology, 26: 13-33,

Davis-Blake, Alison, and JeffreyPfeffer1989 "Just a mirage: The search

for dispositional effects inorganizational research,"Academy of ManagementReview, 14: 385-400,

Development DimensionsInternational1983 Industrial Manager's In-basket,

Pittsburgh, PA: DevelopmentDimensions International,

326/ASQ, June 1993


Diener, Ed, and Robert A.Emmons1984 "The independence of

positive and negative affect."Journal of Personality andSocial Psychology, 47:1105-1117.

Duncker, Karl1945 "On problem-solving."

Psychological Monographs.58: Whole No. 5.

Dunning, David, and Amber L.Story1991 "Depression, realism, and the

overconfidence effect: Arethe sadder wiser whenpredicting future actions andevents?" Journal ofPersonality and SocialPsychology, 61: 521-532.

Fishbein, Martin, and leek Ajzen1975 Belief, Attitude, Intention, and

Behavior: An Introduction toTheory and Research.Reading, MA:Addison-Wesley.

Frederiksen, Norman1966 "Validation of a simulation

technique." OrganizationalBehavior and HumanPerformance, 1: 87-109.

Fried, Rona, and LeonardBerkowitz1979 "Music hath charms . . . and

can influence helpfulness."Journal of Applied SocialPsychology, 70: 314-328.

George, Jennifer M., and ArthurP. Brief1992 "Feeling good—doing good: A

conceptual analysis of themood at work—organizationalspontaneity." PsychologicalBulletin, 112: 310-329.

Golin, Sanford, Francis Terrell,and Barbara Johnson1977 "Depression and the illusion

of control." Journal ofAbnormal Psychology, 86:440-442.

Gouaux, Charles1971 "Induced affective states and

interpersonal attraction."Journal of Personality andSocial Psychology, 20: 37-43.

House, Robert J., Ann Howard,and Gordon Walker1991 "The prediction of managerial

success: A competitive testof the person-situationdebate." Paper presented atthe Annual Meeting of theAcademy of Management,Miami.

Iffaldano, Michelle T., and Paul M.Muchinsky1985 "Job satisfaction and job

performance: A metaanalysis." PsychologicalBulletin, 97: 251-273.

Isen, Alice M., and Robert A.Baron1991 "Positive affect as a factor in

organizational behavior." InBarry M. Staw and Larry L.Cummings (eds.). Research inOrganizational Behavior, 13:1-54. Greenwich, CT: JAIPress.

Isen, Alice M., and Kimberly A.Daubman1984 "The influence of affect on

categorization." Joumal ofPersonality and SocialPsychology, 47: 1206-1217.

Isen, Alice M., Kimberly A.Daubman, and Gary P. Nowicki1987 "Positive affect facilitates

creative problem solving."Journal of Personality andSocial Psychology, 52:1122-1131.

Isen, Alice M., Mitzi M. Johnson,Elizabeth Mertz, and Gregory F.Robinson1985 "The influence of positive

affect on the unusualness ofword associations." Journal ofPersonality and SocialPsychology, 48: 1413-1426.

Isen, Alice M., and Barbara Means1983 "The influence of positive

affect on decision-makingstrategy." Social Cognition, 2:18-31.

Isen, Alice M., Thomas E. Nygren,and Gregory F. Ashby1988 "The influence of positive

affect on the subjective utilityof gains and losses: It's notworth the risk." Journal ofPersonality and SocialPsychology, 55: 710-717.

Isen, Alice M., and Stanley F.Simmonds1978 "The effect of feeling good on

a helping task that isincompatible with goodmood." Social Psychology,41: 346-349.

John, Oliver P., and Richard W.Robins1991 "Accuracy and bias in

self-perception: The role ofindividual differences inself-enhancement andnarcissism." Unpublishedmanuscript. Institute ofPersonality and SocialResearch, University ofCalifornia, Berkeley.

Katz, Daniel, and Ezra Stotland1959 "A preliminary statement to

the theory of attitudestructure and change." InSeymour Koch (ed.).Psychology: A Study of aScience, 3: 423-475. NewYork: McGraw-Hill.

Krzystofiak, Frank, Robert Cardy,and Jerry Newman1988 "Implicit personality and

performance appraisal: Theinfluence of trait inferences onevaluations of behavior."Journal of Applied Psychology,73: 515-521.

Langer, Ellen J.1975 "The illusion of control."

Journal of Personality andSocial Psychology, 32:311-328.

Lazarus, Richard S.1991 Emotion and Adaptation. New

York: Oxford University Press.

Lichtenstein, Sarah, BaruchFischoff, and Lawrence D. Phillips1982 "Calibration of probabilities:

The state of the art to 1980."In Daniel Kahneman, PaulSlovic, and Amos Tversky(eds.). Judgment underUncertainty: Heuristics andBiases: 306-334. Cambridge:Cambridge University Press.

Locke, Edwin A.1976 "The nature and causes of job

satisfaction." In Marvin D.Dunnette (ed.). Handbook ofIndustrial and OrganizationalPsychology: 1297-1349.Chicago: Rand McNally.

Mackie, Diane M., and Leila T.Worth1991 "Feeling good, but not

thinking straight: The impactof positive mood onpersuasion." In Joseph P.Forgas (ed.). Emotion andSocial Judgments: 201-219.Oxford: Pergamon.

March, James G., and Herbert A.Simon1958 Organizations. New York:

Wiley.

Martin, David J., Lyn Y.Abramson, and Lauren B. Alloy1984 "The illusion of control for self

and others in depressed andnondepressed collegestudents." Journal ofPersonality and SocialPsychology, 46: 125.-136.

Matlin, Margaret W., and David J.Stang1978 The Pollyanna Principle:

Selectivity in Language,Memory and Thought.Cambridge, MA: Schenkman.

327/ASQ, June 1993

Mednick, S. A.1962 "The associative basis of the

creative process."Psychological Review, 69:220-232.

Miner, Anne S.1987 "Idiosyncratic jobs in

formalized organizations."Administrative ScienceQuarterly, 32: 327-351.

Mintzberg, Henry1973 The Nature of Managerial

Work. New York: iHarper &Row.

Mobley, William H.1982 Employee Turnover: Causes,

Consequences and Control.Reading, MA:Addison-Wesley.

O'Reilly, Charles A., and JenniferA. Chatnian1986 "Organizational commitment

and psychological attachment:The effects of compliance,identification, andinternalization of prosocialbehavior." Journal of AppliedPsychology, 71: 492^99.

Organ, Dennis W.1988 Organizational Citizenship

Behavior: The Good SoldierSyndrome. Lexington, MA:Lexington Books.

Petty, Richard E., and John T.Cacioppo1986 Communication and

Persuasion: Central andPeripheral Routes to AttitudeChange. New York:Springer-Verlag.

Rafaeli, Anat, and Robert I. Sutton1989 "The expression of emotion in

organizational life." In BarryM. Staw and Larry L.Cummings (eds.). Research inOrganizational Behavior, 11:1-42. Greenwich, CT: JAIPress.

Scheier, Michael F., Jagdish K.Weintraub, and Charles S. Carver1986 "Coping with stress:

Divergent strategies ofoptimists and pessimists."Journal of Personality andSocial Psychology, 48:1162-1172.

Schwarz, Norbert, Herbert Bless,and Gerd Bohner1991 "Mood and persuasion:

Affective states influence theprocessing of persuasivecommunication." In Mark P.Zanna (ed.). Advances inExperimental SocialPsychology, 24: 161-199.New York: Academic Press.

Selignfian, Martin E. P., and PeterSchulman1986 "Explanatory style as a

predictor of productivity andquitting among life insurancesales agents." Journal ofPersonality and SocialPsychology, 50: 832-838.

Smith, Patricia C, Lome M.Kendall, and Charles L. Hulin1969 The Measurement of

Satisfaction in Work andRetirement: A Strategy for theStudy of Attitudes. Chicago:Rand McNally.

Smither, James W., HillarieCollins, and Richard Buda1989 "When ratee satisfaction

influences performanceevaluations: A case of illusorycorrelation." Journal ofApplied Psychology, 74:599-605.

Staw, Barry M.1986 "Organizational psychology

and the pursuit of thehappy/productive worker."California ManagementReview, 28 (4): 40-53.

Staw, Barry M., Nancy E. Bell, andJohn A. Clausen1986 "The dispositional approach to

job attitudes: A lifetimelongitudinal test."Administrative ScienceOuarterly, 31: 56-77.

Staw, Barry M., and Richard D.Boettger1990 "Task revision: A neglected

form of work performance."Academy of ManagementJournal, 33: 534-559.

Staw, Barry M., and Jerry Ross1985 "Stability in the midst of

change: A dispositionalapproach to job attitudes."Journal of Applied Psychology,70: 469-480.

Staw, Barry M., Robert I. Sutton,and Lisa H. Pelled1993 "Employee positive emotion

and favorable outcomes at theworkplace." OrganizationScience (in press).

Sutton, Robert I.1991 "Maintaining norms about

expressed emotions: Thecase of bill collectors."Administrative ScienceOuarterly, 36: 245-268.

Sweeney, Paul D., KarenAnderson, and Scott Bailey1986 "Attributional style in

depression: A meta-analyticreview." Journal ofPersonality and SocialPsychology, 50: 974-991.

Tabachnik, Naomi, JenniferCrocker, and Lauren B. Alloy1983 "Depression, social

comparison, and thefalse-consensus effect."Joumal of Personality andSocial Psychology, 45:688-699.

Taylor, Shelley E,, and JonathonD. Brown1988 "Illusion and well-being: A

social psychologicalperspective on mentalhealth." PsychologicalBulletin, 103: 193-210.

Tellegen, Auke1982 Brief Manual of the

Differential PersonalityOuestionnaire. Minneapolis:University of MinnesotaPress.

1985 "Structures of mood andpersonality and their relevanceto assessing anxiety, with anemphasis on self-report." InA. H. Tuma and J. D. Maser(eds.). Anxiety and the AnxietyDisorders: 681-706. Hillsdale,NJ: Eribaum.

Thornton, George C, and WilliamC. Byham1982 Assessment Centers and

Managerial Performance. NewYork: Academic Press.

Watson, David1988 "The vicissitudes of mood

measurement: Effects ofvarying descriptors, timeframes, and response formatson measures of positive andnegative affect." Journal ofPersonality and SocialPsychology, 55: 128-141.

Watson, David, Lee A. Clark, andGreg Carey1988 "Positive and negative

affectivity and their relation toanxiety and depressivedisorders." Journal ofAbnormal Psychology, 97:346-353.

Watson, David, and Auke Tellegen1985 "Toward a consensual

structure of mood."Psychological Bulletin, 98:219-235.

328/ASO, June 1993


Ul 0)

II

oCO O

> (A

a)

Cor

co

'(/)

deci

co<3

eleg

ocfl)

Eot<uQ.

ing

(0

Eco

'ui'cj0)O

c

E•P

_oCOD

*(J0)

T3

U

ita

4...

(D

. ^CD

X

T3

CO(U

uc

; : O

5 °>o. 2 c COCO 15 ^

10

.2

CN CN . . . .OOOJlfl* COCDoooocomoocNicN rooooooooooo r--o

CD CNo o CD • * 4 - — i n inOOOO'>3-CNOOCN<—oooooooooo

CM CS] —o o c N ac N a m O ^

OOOCNOOoooooo O O CD O

f^ f^ LO C ^ ^^ C^ ^^ CO 00 00 ^^ (ooppcot-oo^p M;(

• ._i ._i • • (Q

T-S _ . „ „ _ :O O '— — CN CO »— — ^ »— 00

pppp^ppppp pp' ~ ' - '— \~ ' - c

rsi CN .—. mO O CD in CD • *OOOOOCOOt— 'tCN CNCJ)oooooooooo --O

t-.. CN]O O CD — I

cN

CN]O O CN CD —oooomc»3OincN (J)oooooooooo t-o

•ac

2u"Sto•o

S o

o £O OiJ

viCL 2

* 3=. . (1)

58

iV 2 V

329/ASQ, June 1993

ro

OC

I

C/}

o

0>Oc(0

pli

Eoo

en'sc o

<DU

c

E

I(DQ.

cID

io•coQ.

CUa.

ter

c

m

c0]

.^ E(A Q)

t - roU)c(U

•si

o o

o ro

a

o

art

Q.

CO

c

filfillro <o ^ <»" E9!tE 9 ti_ E 11)

CN 'J- • , . —oO'-O'-ro'*'—cocM rocoOO"-"—O'S-'-"—coco m o

C M t ( D CNCOt-'-co mo

ID ^ .o o o c o * c N O O n.

. - c N O O iS

0)00+-^oocoOO"—

ro g^cNokiro ?

^ . . a00 T)" , , , • , , •oocD<-cDooocooooo oooo>—' -o in ' - '—oco > — • —

? o r o m r r c N c N r - L n cs'>^

I CO. _ ~ ~ l _ I"-O"—roco'<tcNOoro r-cx3

' — ' ~^ ' -^ ' -^ ' -^ m ' •

>—ocorooco>—omoOOCNO'^T'd'O'— OCO

CD-ac

, CD

CO CM — — .—.• ^ -OO^LOCOOfCDOOO

oS?C 3 CD

^ §3

- £ CDp o

en i-L (_)

330/ASQ. June 1993


Table A,3

Regressions of Control Variables and Affect on Ratings of ManagerialDimensions (/V = 111)*

Ratings on managerial dimensionsManagerial Overalldimension managerial

Variable scale potential

,004*^(.001),02

(.03),10

(.15),03

(.03). 27 -

(.10)5,47—

,17

• p < ,05; "p < ,01; •~p < ,001; two-tailed tests,• Unstandardized coefficients are reported, with standard errors in parenthe-

ses.

GMAT

Age

Gender

Years of workexperience

Affect

F-ratioAdjusted R^

,002-(.001).02

(.02).28 -

(.11).01

(.02).15*

(.07)4,59—

,14

331/ASQ, June 1993

Affect and Managerial Perfornnance: A Test of Hypotheses Barry...

Documents

Transcript of Affect and Managerial Perfornnance: A Test of Hypotheses Barry...